Driver Drowsiness Detection Based on HOG and Gabor Features

Imranul HAQ, Sheharyar, Jia-cai ZHANG

Abstract


Drowsiness detection system can be very useful for driving behavior monitoring and operator attentiveness detection. Traffic accidents can be seen in every part of the world. To avoid accidents, we need a system to continuously monitor the drowsiness level and warn the driver if he/she feels drowsy. Driver drowsiness detection from facial video analysis is the scope of this paper. In this paper we proposed a non-intrusive drowsiness detection system which utilizes different facial features of driver to detect and alert drowsy driver. The classifier output is then given to Alert unit to give warning to the driver if danger exists. By implementing this simple method, the system achieved 91 % accuracy on NN and 88 % accuracy on SVM using public NTHU-DDD dataset outperforming other methods using Expensive hardware to reach the same goal.

Keywords


Histogram of Oriented Gradients (HOG), Gabor, Driver drowsiness, Vision based measurement, Support Vector Machine (SVM) and Neural Networks (NN)


DOI
10.12783/dtssehs/iceme2019/29597

Full Text:

PDF